A little creativity, a lot of code.

Hi, I'm Elisa, a Computer Engineering student at HVL in Bergen, building thoughtful websites, useful software and intelligent systems.

Selected work

A real website for a real organization, and the AI systems I'm proudest of. Each one says plainly what it is, and how far along it is.

Experiments & learning

Smaller builds, prototypes and things in progress. Honest labels included.

  • Assumption Hunter

    Evaluated

    Finds the hidden assumptions a codebase relies on, each tied to a file and the risk if it is false.

  • TeachBack AI

    Prototype

    Grades how well someone explains a concept, against fixed rubrics. The engine only, no interface yet.

    TypeScript · Pixel Forge AI HackathonSource (opens in a new tab)
  • Language Agent

    In progress

    My own Norwegian tutor on Telegram. Spaced repetition decides what comes back; Claude writes and grades the lesson.

    Python · SQLite · ClaudeSource (opens in a new tab)
  • SafeHer

    In development

    A discreet personal-safety app with a silent SOS, co-founded with Alissa Ruya Aksel at HVL.

    Flutter · Node.js · PostgreSQL
  • ElisaTech chatbot

    Live

    A lead-capture chatbot that answers in Norwegian, English and Turkish and hands every lead to the owner.

    n8n · Google Sheets

What I can build for you

For nonprofits, organizations and small businesses. Clear scope, honest estimates, and a site or tool you understand when it's handed over.

  1. Discover
  2. Plan
  3. Design
  4. Build
  5. Review
  6. Launch
Tell me about your project

How I think through a project

One real example, from Assumption Hunter: what went in, and what I learned when the numbers came back.

  1. Problem

    Teams inherit code full of assumptions nobody wrote down, until one breaks in production.

  2. Constraints

    Every finding has to point to a real file and a real risk. Vague advice does not count.

  3. Decisions

    Start with a single-prompt baseline, then add an evidence checker only where it can be measured.

  4. Build

    A two-stage pipeline in Python: an analyzer proposes, a checker verifies against the cited file.

  5. Tests

    Ten small projects with known answers, each with a test proving its counterexample really fails.

  6. Learning

    The checker approved almost everything. A severity filter cut the noise by 22% without losing a single real hit.

Hi, I'm Elisa

I study Computer Engineering at HVL in Bergen. I'm drawn to problems where software has to be careful: official letters, personal safety, systems that act on their own. And I like making things that feel good to use.

I speak Norwegian, English and Turkish, and I build for people in all three. Most of what I know I learned by shipping something, testing it like someone wants it to break, and writing down what surprised me.

Tools I've used in these projects

Languages
TypeScriptPythonDartSQL
Interfaces
Next.jsReactTailwind CSSFlutter
Backends & data
FastAPINode.jsPostgreSQLSupabaseFirestoreSQLite
AI
Gemini on Vertex AIGoogle ADKClaude APIn8n
Shipping & quality
VercelCloud RunGitHubpytestVitestPlaywright

Have something lovely in mind? Let's build it thoughtfully.

A website, a tool, an internship or just a question: write a few lines and I'll get back to you.

Your message goes to my inbox through Formspree and is only used to reply to you.